Chapter 4: Morphology

possessive compounds or bahuvrihi compounds, a term which comes from the ancient 
Indian language Sanskrit and exemplifies the phenomenon itself, as it literally expresses the 
notion of „having a lot of rice‟ but means „rich man‟. The third type of compound is also often 
referred to by a Sanskrit term, dvandva, meaning „pair‟. Dvandva compounds are compounds 

---

18 
in which there is no modifier-head relation, but both constituents are considered as heads on a 
par, e.g. study-bedroom, singer-songwriter, bitter-sweet or deaf-mute. They are also known as 
copulative compounds when they denote the sum of the two meanings (cf. bitter-sweet), or 
as appositional compounds when they combine two different descriptions of the referent 
(singer-songwriter). Unlike determinative compounds, dvandva compounds are typically 
stressed on both elements of the pair. As is shown in Table 4.6, the three major types of 
compounds can also be differentiated in terms of the logical relations between the constituents 
and the compound: 
Table 4.6: Survey of types of compounds differentiated by internal morphological and 
semantic structure 
type 
internal structure 
logical relation (A = first 
constituent; B = second con-
stituent; AB = compound) 
example and 
paraphrase 
determinative 
compounds 
modifier-head structure, 
endocentric (head is part 
of compound) 
AB is a type of B
mousemat 
„a mousemat is a type 
of mat‟ 
bahuvrihi 
compounds 
exocentric (head is 
found outside 
compound) 
AB is neither A nor B but a 
type of C 
egghead 
„an egghead is neither 
a type of egg nor a 
type of head but a type 
of person‟ 
dvandva 
compounds, 
either copulative 
or appositive 
two-headed structure, 
endocentric 
AB is both A and B 
singer-songwriter 
„a person who is both 
a singer and a 
songwriter‟ 
A special challenge for the analysis of morphological structure arises in synthetic compounds 
of the nominal types theatre-goer and shareholding and the adjectival types eye-catching and 
dark-haired, all of which involve verbal elements and bound lexical morphemes. The problem 
concerns the branching in binary immediate constituents and the allocation of modifier and 
head roles. In all four cases, an analysis in terms of a compound consisting of a simple 
modifier (theatre, share, eye and dark) and a suffixed head is ruled out, as the potentials heads 
goer, holding, catching and haired are at least doubtful with regard to their status as existing 
lexemes. Analyses in terms of suffixations with complex modifiers (to theatre-go + -er, to 
sharehold + ing, to eye-catch + -ing and dark-hair + -ed) are equally unsatisfactory on the 
same grounds that the potential bases do not exist. In these cases, and also in those numerable 
ones where a compound analysis seems at least possible, for instance for bus driver, it may 
seem advisable to argue that compounding and suffixation take place at the same time, so to 
speak, and to regard these lexemes as synthetic compounds formed by compressing major 

---

19 
components of sentences into one word (cf. theatre-goer  „someone who goes to the 
theatre‟).
Further, somewhat less typical classes of compounds include so-called phrase compounds 
(e.g. father-in-law, rough-and-ready, man-in-the-street, good-for-nothing) and particle 
compounds derived from phrasal verbs (take-away, breakthrough, handout, take-off), which 
present a serious problem for modifier-head analysis. Neoclassical compounds are formations 
that also combine two concepts in a manner very similar to compounds, but these are not 
encoded by free lexical morphemes, but rather by bound forms derived from Greek and, less 
frequently, Latin. Examples of these somewhat learned and often technical words include 
democrat, photograph, biography, technology and microscope. 
The types of semantic structures and internal relations that can be realized by compounds are 
virtually unlimited. Nevertheless, some tendencies concerning particularly frequent types can 
be identified. The examples of root compounds featuring chair as head have already given 
you a glimpse of some of the most dominant relations. More examples are provided in Table 
4.7:
Table 4.7: Frequent semantic relations in root compounds 
relation 
examples 
FUNCTION
gunpowder, breadbasket, toothbrush